The selective expression in tumor cells of non-human genes that can produce toxic compounds from non-toxic compounds is being considered for the treatment of various solid tumors that are refractory to existing chemotherapeutics agents. A clinical trial at the NCI as begun to evaluate the potential for activating ganciclovir in inoperable brain tumors that have been transduced with the herpes simplex virus thymidine kinase (HSV-TK) gene. However, we believe that the inefficient delivery of the HSV-TK gene to human tumor cells in vivo together with inadequate ability to kill neighboring cells that are not transduced with the HSV-TK gene represent two major practical hurdles in this approach to the gene therapy of cancer. Because of these problems with the use of HSV-TK to activate compounds, we have developed a strategy to deliver toxic compounds that will not be trapped in the cell in which they are formed. We have utilized the substrate differences between human and E. coli purine nucleoside phosphorylase (PNP) to activate non-toxic prodrugs to toxic purine bases. E. coli PNP recognizes adenine-containing nucleosides as substrates whereas human PNP does not. Preliminary studies have shown that expression of E. coli PNP in less than 1% of human cancer cells in culture leads to the death of virtually all bystander cells after treatment with the prodrug 6-methylpurine-2'-deoxyriboside (MeP-dR). MeP- dR is a relatively nontoxic deoxyadenosine nucleoside analog that can be converted to MeP, a toxic purine base, by E. coli PNP, but not by human PNP. In addition, our preliminary studies indicate that some of the agents that could be created in this manner kill both replicating and non- replicating cells, which further distinguishes our approach for the treatment of solid tumors from most of the currently used antitumor agents. The longterm goal of this NCDDG is to develop a cancer treatment strategy based on the selective expression of E. coli PNP in tumor cells to activate nontoxic purine analog prodrugs. This NCDDG is composed of 4 programs and one core. The objectives of the components of this NCDDG are: 1) Molecular Biology Program to develop procedures to selectively transfect or transduce the E. coli PNP gene into tumor cells of whole animals; 2) Biochemistry Program to fully characterize the biochemical pharmacology of the purine nucleoside analogs and their respective bases to aid in the rational design of new prodrugs; 3) Chemistry Program to design and synthesize nontoxic purine nucleoside prodrugs that will be converted into toxic purine bases in tumor cells that express the E. coli PNP gene; 4) X-ray Crystallography Program to determine the structure of E. coli PNP and to compare it with the structure of mammalian enzymes to aid in the rational design of new prodrugs; and 5) Chemotherapy Core to evaluate the antitumor activity of the prodrugs developed in this NCDDG in relevant animal tumor models developed with the aid of the Molecular Biology Program.